Clinical background
Venovenous ECMO for severe acute respiratory failure is resource-intensive and carries a substantial risk of complications. The decision to start ECMO is often made under time pressure in a patient who has not responded to conventional ventilation, and clinical judgements about prognosis vary widely between assessors. The RESP score was developed to structure this assessment using variables available immediately before cannulation. The score was, however, derived among patients already selected for ECMO, and it predicts survival given that treatment is started. It therefore cannot in itself determine whether a patient facing the decision will benefit from ECMO compared with continued conventional treatment [5].
Calculating the RESP score
The RESP score is a weighted sum of twelve pre-ECMO variables:
where each variable is assigned a weight as follows:
| Variable | Category | Points |
|---|---|---|
| Age | 18–49 | 0 |
| 50–59 | −2 | |
| ≥60 | −3 | |
| Immunocompromised | No | 0 |
| Yes | −2 | |
| Mechanical ventilation before ECMO | <48 hours | 3 |
| 48 h to 7 days | 1 | |
| >7 days | 0 | |
| Acute respiratory diagnosis | Viral pneumonia | 3 |
| Bacterial pneumonia | 3 | |
| Asthma | 11 | |
| Trauma and burns | 3 | |
| Aspiration pneumonitis | 5 | |
| Other acute respiratory diagnoses | 1 | |
| Non-respiratory/chronic respiratory diagnoses | 0 | |
| Central nervous system dysfunction | No | 0 |
| Yes | −7 | |
| Acute associated (non-pulmonary) infection | No | 0 |
| Yes | −3 | |
| Neuromuscular blockade before ECMO | No | 0 |
| Yes | 1 | |
| Inhaled nitric oxide before ECMO | No | 0 |
| Yes | −1 | |
| Bicarbonate infusion before ECMO | No | 0 |
| Yes | −2 | |
| Cardiac arrest before ECMO | No | 0 |
| Yes | −2 | |
| PaCO₂ ≥75 mmHg | No | 0 |
| Yes | −1 | |
| Peak inspiratory pressure ≥42 cmH₂O | No | 0 |
| Yes | −1 |
The score ranges from −22 to +15. The variables reflect both the patient's chronic status (age, immunosuppression, diagnostic group) and the severity of the acute illness together with the rescue treatments already tried (bicarbonate, nitric oxide, neuromuscular blockade, cardiac arrest).
The derivation cohort consisted of 2,355 adult patients with severe acute respiratory failure treated with ECMO between 2000 and 2012, extracted from the international ELSO registry. In all, 1,338 patients (57%) were discharged alive from hospital. Multivariable logistic regression with bootstrap methods was used to identify variables independently associated with hospital survival [1]. The model was validated internally and externally in 140 patients, with a c-statistic of 0.92 (95% CI 0.89–0.97) in the external cohort [1].
Interpretation in practice
The RESP score translates into five risk classes with the associated hospital survival:
| Class | Score | Expected survival | Clinical interpretation |
|---|---|---|---|
| I | ≥6 | 92% | Very good prognosis. ECMO is reasonably justified if the other criteria are met. |
| II | 3–5 | 76% | Good prognosis. Treatment should be offered. |
| III | −1 to 2 | 57% | Intermediate prognosis. The decision must be weighed against comorbidity, the expected time to lung recovery and the patient's own wishes. |
| IV | −5 to −2 | 33% | Poor prognosis. ECMO may be considered in selected cases, but the expected benefit is limited. |
| V | ≤−6 | 18% | Very poor prognosis. ECMO is generally not justified unless there are particular reasons (a bridge to transplantation, a reversible cause). |
The survival figures derive from the derivation cohort and should be interpreted as population-based estimates, not as individual predictions. No single score constitutes an absolute threshold for withholding ECMO. A patient in class V may still survive, and a patient in class I may die of complications related to the ECMO treatment itself. The score should always be placed in its clinical context: the reversibility of the underlying disease, the waiting time for lung transplantation, patient autonomy and the intensive care experience of the centre.
Validation and performance
The derivation study achieved a c-statistic of 0.74 (95% CI 0.72–0.76) in the development cohort and 0.92 (95% CI 0.89–0.97) in the external validation cohort of 140 patients [1]. The high figure in the external cohort should be interpreted with caution given its small size and the possibility that the cohort was selected.
In subsequent external validations, discrimination has varied considerably. In a German single-centre study of 108 ARDS patients on venovenous ECMO (2010–2015), the AUC was 0.64 (p = 0.012), with acceptable calibration by the Hosmer–Lemeshow test (χ² = 8.4, p = 0.395) [2]. In a Chinese single-centre study of 23 patients, the AUC was 0.835 (95% CI 0.659–1.010) with an optimal cut-off at risk class 3.5 and a specificity of 84.6% [3]. In a Korean cohort of 50 patients (2012–2014), the AUC was 0.79 (95% CI 0.65–0.89), with a cut-off at ≤−1 giving 91.3% specificity and 66.7% sensitivity for in-hospital mortality [4].
A systematic review from 2023 identified 58 prognostic models for ECMO patients and 225 external validations [5]. The RESP score was one of the most extensively validated models (18 external validations). The pooled c-statistic for RESP lay between 0.66 and 0.70, that is, moderate discrimination and lower than in the derivation cohort. Almost all the models tended to underestimate mortality (O:E > 1), meaning that the expected survival according to the score systematically overestimates the actual figure. Only 1 of the 58 models met the PROBAST criteria for a low risk of bias [5].
Limitations
The RESP score applies only to venovenous ECMO for acute respiratory failure. It is not validated for venoarterial ECMO, for ECMO as a bridge to transplantation or for ECMO in cardiogenic shock. Patients whose dominant problem is chronic lung disease fall into the category "non-respiratory/chronic respiratory diagnoses" with 0 points, which may give a misleadingly high estimate if the acute component is limited.
The most important limitation is that the score was derived among patients already selected for ECMO. It therefore predicts survival given that treatment is started, but cannot answer whether a patient facing the decision will benefit more from ECMO than from continued conventional treatment [5]. Using the score as the sole decision aid for or against ECMO is therefore not justified.
The diagnosis variable is crudely categorised. Asthma scores 11 points, which in practice places almost all asthma patients in class I regardless of other factors. This reflects the excellent prognosis of asthma on ECMO but may mask differences in severity within the group. Mechanical ventilation for >7 days before ECMO scores 0 points, which penalises patients with prolonged ventilation, but the 7-day boundary is arbitrary and reflects the distribution in the derivation cohort rather than a physiological threshold.
Calibration is a concern. The systematic review found that models tend to underestimate mortality [5], meaning that a patient expected by the RESP score to have 76% survival may in practice have a lower probability. This is particularly marked in cohorts with a high proportion of immunocompromised patients or with sepsis-dominated ARDS.
References
- Schmidt M, Bailey M, Sheldrake J, et al. Predicting survival after extracorporeal membrane oxygenation for severe acute respiratory failure. The Respiratory Extracorporeal Membrane Oxygenation Survival Prediction (RESP) score. Am J Respir Crit Care Med. 2014;189(11):1374–1382. PMID: 24693864
- Hilder M, Herbstreit F, Adamzik M, et al. Comparison of mortality prediction models in acute respiratory distress syndrome undergoing extracorporeal membrane oxygenation and development of a novel prediction score: the PREdiction of Survival on ECMO Therapy-Score (PRESET-Score). Crit Care. 2017;21:301. PMID: 29233160
- Huang L, Li T, Xu L, et al. Performance of Multiple Risk Assessment Tools to Predict Mortality for Adult Respiratory Distress Syndrome with Extracorporeal Membrane Oxygenation Therapy: An External Validation Study Based on Chinese Single-center Data. Chin Med J (Engl). 2016;129(14):1688–1695. PMID: 27411456
- Lee S, Yeo HJ, Yoon SH, et al. Validity of Outcome Prediction Scoring Systems in Korean Patients with Severe Adult Respiratory Distress Syndrome Receiving Extracorporeal Membrane Oxygenation Therapy. J Korean Med Sci. 2016;31(6):932–938. PMID: 27247503
- Pladet LCA, Barten JMM, Vernooij LM, et al. Prognostic models for mortality risk in patients requiring ECMO. Intensive Care Med. 2023;49(2):131–141. PMID: 36600027
